flykhan/maths-cs-ai-compendium-zh
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Deployed 2536c93 with MkDocs version: 1.6.1

Browse Source
This commit is contained in:
flykhan
2026-05-03 11:47:30 +08:00
commit 1ff86b66fc
418 changed files with 617336 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
images/actor_critic.svg
+45
View File
@@ -0,0 +1,45 @@
<svg width="700" height="300" xmlns="http://www.w3.org/2000/svg">
<defs>
<marker id="ac-arrow" markerWidth="8" markerHeight="6" refX="8" refY="3" orient="auto">
<path d="M0,0 L8,3 L0,6" fill="none" stroke="#555" stroke-width="1"/>
</marker>
</defs>
<text x="350" y="22" fill="#333" font-size="14" font-weight="bold" text-anchor="middle">Actor-Critic Architecture</text>
<!-- Shared input -->
<rect x="280" y="240" width="140" height="30" rx="6" fill="#f5f5f5" stroke="#333" stroke-width="1.5"/>
<text x="350" y="260" fill="#333" font-size="11" font-weight="bold" text-anchor="middle">State s_t</text>
<!-- Shared backbone (optional) -->
<line x1="320" y1="240" x2="200" y2="195" stroke="#555" stroke-width="1.5" marker-end="url(#ac-arrow)"/>
<line x1="380" y1="240" x2="500" y2="195" stroke="#555" stroke-width="1.5" marker-end="url(#ac-arrow)"/>
<!-- Actor head (left) -->
<rect x="115" y="140" width="170" height="55" rx="8" fill="#3498db" opacity="0.15" stroke="#3498db" stroke-width="2"/>
<text x="200" y="162" fill="#3498db" font-size="13" font-weight="bold" text-anchor="middle">Actor (Policy)</text>
<text x="200" y="180" fill="#3498db" font-size="11" text-anchor="middle">pi(a|s; theta)</text>
<!-- Critic head (right) -->
<rect x="415" y="140" width="170" height="55" rx="8" fill="#e74c3c" opacity="0.15" stroke="#e74c3c" stroke-width="2"/>
<text x="500" y="162" fill="#e74c3c" font-size="13" font-weight="bold" text-anchor="middle">Critic (Value)</text>
<text x="500" y="180" fill="#e74c3c" font-size="11" text-anchor="middle">V(s; phi)</text>
<!-- Actor output -->
<line x1="200" y1="140" x2="200" y2="105" stroke="#3498db" stroke-width="1.5" marker-end="url(#ac-arrow)"/>
<rect x="135" y="75" width="130" height="30" rx="14" fill="#3498db" opacity="0.2" stroke="#3498db" stroke-width="1.5"/>
<text x="200" y="95" fill="#3498db" font-size="11" font-weight="bold" text-anchor="middle">Action a_t</text>
<!-- Critic output -->
<line x1="500" y1="140" x2="500" y2="105" stroke="#e74c3c" stroke-width="1.5" marker-end="url(#ac-arrow)"/>
<rect x="430" y="75" width="140" height="30" rx="14" fill="#e74c3c" opacity="0.2" stroke="#e74c3c" stroke-width="1.5"/>
<text x="500" y="95" fill="#e74c3c" font-size="11" font-weight="bold" text-anchor="middle">Value estimate</text>
<!-- Advantage signal from critic to actor -->
<path d="M 430,85 C 380,55 320,55 285,75" fill="none" stroke="#9b59b6" stroke-width="1.8" stroke-dasharray="5,3" marker-end="url(#ac-arrow)"/>
<text x="350" y="48" fill="#9b59b6" font-size="11" font-weight="bold" text-anchor="middle">advantage = r + gamma*V(s') - V(s)</text>
<text x="350" y="64" fill="#9b59b6" font-size="10" text-anchor="middle">(guides policy update)</text>
<!-- Summary -->
<rect x="120" y="280" width="460" height="18" rx="4" fill="#f5f5f5" stroke="#333" stroke-width="0.8"/>
<text x="350" y="293" fill="#333" font-size="10" text-anchor="middle">Actor decides what to do. Critic evaluates how good the decision was.</text>
</svg>
Side by Side

After

Width:  |  Height:  |  Size: 3.0 KiB